DOCTORA L T H E S I S
Department of Civil, Mining and Environmental Engineering Division of Architecture and Infrastructure
Biofiltration Technologies for
Stormwater Quality Treatment
Godecke-Tobias Blecken
ISSN: 1402-1544 ISBN 978-91-7439-132-9Luleå University of Technology 2010
Godeck
e-T
obias
Bleck
en
Biofiltration
Technolo
gies
for
Stor
mw
ater
Quality
T
reatment
Biofiltration Technologies
for Stormwater Quality Treatment
Godecke-Tobias Blecken
Urban Water
Division of Architecture and Infrastructure
Department of Civil, Mining and Environmental Engineering Luleå University of Technology
SE-971 87 LULEÅ SWEDEN
Printed by Universitetstryckeriet, Luleå 2010 ISSN: 1402-1544
ISBN 978-91-7439-132-9 Luleå 2010
ACKNOWLEDGEMENTS
The experiments that were conducted at Luleå University of Technology were financially supported by Åke och Greta Lissheds stiftelse, J. Gust. Richert stiftelse and Luleå
University of Technology. My participation in conferences and my research stay at
Monash University in 2007 were supported by Seth M Kempes Stipendiefond,
Wallenbergsstiftelsen - Jubileumsanslaget - and Stiftelsen Futura. Sincere thanks are
extended for all that support. I also would like to thank the Peter Stahre stipendium committee for acknowledging the scientific value and practical applicability of this research.
This work would not have been possible without the help of Monica Olofsson and Kerstin Nordqvist with the laboratory work. Huge thanks are also due to all other colleagues in the Urban Water group who contributed in one way or another to my doctoral efforts, by creating the stimulating work atmosphere at the research group at LTU, including discussions about work and private life during lots of coffee breaks. I would like to thank Tone Muthanna for providing insights regarding her own PhD studies and for being an inspiring author of several papers. Thanks also to my co-author Jiri Marsalek, for sharing his extraordinary experience which considerably improved this work.
During 2007 I spent six months at Monash University in Melbourne, Australia. I would like to thank everyone there who helped to make my stay there a good experience. Yaron Zinger is especially acknowledged for conducting lots of experiments, sharing his data and always being a serious and competent discussion partner and co-author. I do not know which of my co-supervisors, Ana Deletic and Tim Fletcher, I should name first here, so I have put your names in alphabetical order. Both of you helped me greatly by enabling the productive collaboration with FAWB at Monash University. I am very grateful that you always willingly shared your experience, ideas, comments, data and (especially) your scarce time with me. I have learned a lot from you. Thank you!
To my main supervisor Maria Viklander, many thanks for giving me the opportunity to undertake this doctoral research. Thank you for all your time, encouragement and guidance through all the obstacles on the way to presenting this thesis. I also would like to thank you for allowing me to take parental leave whenever I asked.
Finally, I am grateful to have such a wonderful family. Thanks to all of you in Germany for regularly coming up to Luleå – or wheresoever we are. Henneke, Thies and Ehler, without you the last years would have been much less exciting. With your energy you three helped me to leave work behind every evening. Inga, thanks for taking me with you up to Luleå - otherwise there would not have been any thesis. Thanks for being my best friend. You are the perfect one for me. I am very much looking forward to spending more time with you again soon.
ABSTRACT
Due to high runoff volumes and peak flows, and significant contamination with (inter alia) sediment, metals, nutrients, polycyclic aromatic hydrocarbons and salt, urban stormwater is a major cause of degradation of urban water ways. Since current urban drainage systems, which heavily rely on piped sewer networks, may not be sustainable, attempts are being made to develop and refine sustainable urban drainage solutions, notably in Water Sensitive Urban Design (WSUD) and Low Impact Development (LID) concepts. Promising systems recommended for application in both WSUD and LID are stormwater biofilters (also known as bioretention systems or rain gardens) using vegetated filter media. Besides their capacity to attenuate flows and minimise runoff volumes, stormwater biofilters have proven efficacy for enhancing effluent water quality. Furthermore, they can be aesthetically pleasingly integrated even in dense urban environments. However, there are still gaps in our knowledge of the variability of biofilters’ pollutant removal performance, and the factors that affect their performance.
In the studies this thesis is based upon, the effects of various ambient factors, stormwater characteristics and modifications of filter design on the removal of metals, nutrients and total suspended solids (TSS) in biofilters, and pollutant pathways through them, have been investigated. For these purposes, standard biofilters and variants equipped with a submerged zone, a carbon source and different filter materials were exposed to varying temperatures and dry periods, dosed with stormwater and snowmelt, and the inflow and outflow concentrations of the pollutants were measured.
Although removal percentages were consistently high (>70%), demonstrating that biofilters can reliably treat stormwater, the results show that metal outflow concentrations may vary widely depending on the biofilter design and the ambient conditions. Prolonged drying especially impaired their removal efficiency, but variations in temperature and filter media variations had little effect on metal removal rates. The adverse effects of drying could be mitigated by using a submerged zone, and thus providing a more constant moisture regime in the filters between storm events. Combined with embedded organic matter, the submerged zone especially significantly enhances Cu removal, helping to meet outflow target concentrations. Similarly, installing a mulch layer on top of the filter provides additional sorption capacity, hence metals do not ingress far into the filter and are mainly trapped on/in the top layer by sorption processes and/or mechanical trapping associated with TSS. This leads to significant metal accumulation, which facilitates biofilter maintenance since scraping off the top layer removes high proportions of previously accumulated metals, thus delaying the need to replace the whole filter media. However, removal of accumulated pollutants from the filter media is crucial for successful long-term performance of the filters to ensure that no pollutant breakthrough occurs.
Nitrogen removal was found to be more variable than metal removal, and to be adversely affected by temperature increases, leading to high nitrogen leaching in warm temperatures. Phosphorus removal rates were consistently high, since most phosphorus was particle-bound and thus trapped together with TSS. However, in initial stages phosphorus was washed out from the filter media, indicating that filter media that do not have high levels of labile phosphorus should be used to avoid high effluent concentrations.
Given that most outflow concentrations were far lower than those in the stormwater, biofilters are appropriate stormwater treatment systems. Dependent on the ambient conditions, the target pollutants and the sensitivity of the recipient, adaptation of the filter design is recommended. Further work is required to investigate the winter performance and improve the reliability of nitrogen removal, which is highly variable.
SAMMANFATTNING
Dagvatten är en viktig orsak till ekologiska försämringar av urbana vattendrag p.g.a. stora avrinningsvolymer, och höga flöden samt en tillförsel av diverse föroreningar, t.ex. sediment, tungmetaller, näringsämnen, polycykliska aromatiska kolväten och salt. Dagvattenhanteringen har länge varit fokuserad enbart på att leda bort vattnet i rörledningar utan att hänsyn har tagits till retention av stora flöden eller till vattenkvalitén. På grund av dessa problem har utvecklingen av uthålliga dagvattensystem blivit allt viktigare och koncept som Lokalt
Omhändertagande av Dagvatten (LOD), Water Sensitive Urban Design (WSUD) och Low Impact Development (LID) har utvecklats. En uthållig lösning inom dessa koncept är
dagvattenbiofiltrering.
Dagvattenbiofilter är infiltrationsbäddar med växter där dagvattnet infiltrerar och renas av växterna och filtermaterialet. De har en god förmåga att fördröja stora flöden samt att reducera föroreningar i dagvattnet innan det släpps ut till recipienten. Dessutom är det en estetisk och naturnära teknik som mycket väl kan integreras arkitektoniskt i både nya och befintliga stadsmiljöer. Dock saknas det fortfarande mycket kunskap om de processer som styr reningsförmågan samt hur de påverkas av varierande omgivningsförhållanden.
I denna avhandling har därför effekterna av olika omgivningsfaktorer, dagvattenegenskaper och design av biofilter på reningen av metaller, näringsämnen och sediment undersökts. För att undersöka detta har biofilter, som delvis försetts med olika filtermaterial eller en vattenmättad zon, till dels kombinerad med en kolkälla, och utsatts för olika temperaturer och torra perioder. Biofiltren har bevattnats med dagvatten eller smältvatten. Prover har tagits på ingående och utgående vatten och föroreningskoncentrationerna har analyserats.
Trots att reduktionsförmågan av metaller var hög (>70%), vilket bekräftar att biofiltren har förmågan att effektivt rena dagvattnet, visar resultaten att de utgående metallkoncentrationerna kan variera mycket beroende på utformningen av biofilter och varierande omgivningsfaktorer. Torra perioder som är längre än 3 till 4 veckor minskar metallavskiljningen i biofilter, medan växlande temperaturer och olika filtermaterial hade mindre betydelse för metallreningen. Dock kan en vattenmättad zon i filtermaterialet minimera (Cu och Zn) eller till och med avlägsna (Pb) den negativa effekten av torka med avseende på reningsförmågan. I kombination med en kolkälla kan en vattenmättad zon öka reningseffekten för framför allt Cu (som inte är lika bra i standardutförande av biofilter) på grund av en ökad komplexbildning och partikulärt organiskt material. Sediment, metaller och partikelbundna dagvattenföroreningar hålls tillbaka redan i det översta filterlagret vilket leder till en hög metallackumulation. Detta underlättar filterunderhållet: genom att skrapa och ersätta bara det översta jordlagret kan en hög andel ackumulerade föroreningar tas bort från filtret. Således kan utbyte av det hela filtermaterialet fördröjas.
Kvävereningen var inte lika effektiv som metallreningen. I varma temperaturer (20°C) har kväveutlakning i stället for reduktion observerats. Fosforreningen var dock hög eftersom fosfor var mestadels partikelbunden och blev därför filtrerat tillsammans med sedimentet i det översta filterlagret. I början av biofilterdriften har dock fosforurlakning från filtermaterialet observerats vilket tyder på att det inte ska innehåller höga halter av fosfor för att undvika utlakning från filtret.
Eftersom de flesta föroreningskoncentrationer i det utgående vattnet var betydligt lägre än i dagvattnet är biofilter en uthållig och tillförlitlig teknik för dagvattenrening. Beroende på olika omgivningsfaktorer samt de ekologiska förhållandena i recipienten rekommenderas dock anpassning av filterdesignen. Framtida forskning behövs för att undersöka biofiltrens reningsförmåga under vinterförhållanden och för att förbättra den varierande kvävereningen.
ZUSAMMENFASSUNG
Regenwasserabfluss von versiegelten Flächen ist aufgrund des im Vergleich zu ländlichen und naturnahen Einzugsgebieten hohen und schnellen Oberflächenabflusses und seiner Schadstoff-belastung Hauptgrund für die Verschlechterung der Gewässergüte urbaner Gewässer. Da die derzeitige Regenwasserbewirtschaftung (meist basierend auf Misch- oder Trennkanalisation) nicht nachhaltig ist, besteht ein dringender Bedarf an Neu- und Weiterentwicklungen alter-nativer nachhaltiger Systeme. Ein Beispiel hierfür sind Bodenfilter (eng. stormwater biofilter,
bioretention oder rain garden) zur Speicherung, Behandlung und Versickerung von
Regenwas-ser, die vor etwa 20 Jahren in den USA entwickelt worden sind. Forschungsergebnisse und praktische Erfahrungen damit sind viel versprechend und deuten auf ein hohes Potenzial hin-sichtlich Regenwasserrückhalt und -behandlung hin. Außerdem können sie zu einer Auf-wertung des Stadtbildes beitragen. Allerdings besteht nach wie vor Forschungsbedarf über das Reinigungsvermögen von Biofiltern und die Faktoren, die dieses beeinflussen.
Im Rahmen dieser Dissertation wurde anhand von Laborversuchen der Einfluss wechselnder äußerer Rahmenbedingungen, Regenwassereigenschaften und Anpassungen des Bodenfilter-aufbaus auf die Reinigung des Regenwassers von Schwebstoffen, Schwermetallen und Nähr-stoffen untersucht. Für die Versuche wurden Bodenfilter mit einer wassergesättigten Zone (teilweise in Kombination mit organischem Material) und verschiedenen Filtermaterialien versehen, unterschiedlichen Regen-/Trockenphasen sowie unterschiedlichen Temperaturen ausgesetzt und mit Regen- oder Schneeschmelzwasser bewässert.
Obwohl die Metallkonzentrationen im behandelten Wasser deutlich unter denen des unbe-handelten Regenwassers lagen, variierten die Konzentrationen im Ausflusswasser je nach Fil-termaterial und Umgebungsbedingungen stark. Vor allem längere Trockenperioden ver-schlechterten die Reinigungsleistung bei nachfolgenden Regenereignissen deutlich. Die Re-genwasserreinigung nach Trockenperioden ließ sich durch eine wassergesättigte Zone im Fil-ter deutlich verbessern. Diese wassergesättigte Zone steigert außerdem maßgeblich die Kup-ferreinigung, wenn sie mit in den Filter eingebettetem organischem Material kombiniert wurde. Gleiches galt für eine Mulchschicht auf dem Filter, aufgrund der Bereitstellung zusätz-licher Sorptionskapazität. Temperaturunterschiede beeinflussten die Metallreinigung wenig. Metalle drangen im Allgemeinen nicht tief in das Filtermaterial ein, sondern wurden in den oberen Filterschichten zurückgehalten. Hierdurch entstand eine erhebliche Ablagerung von Schwermetallen im obersten Filterbereich, welche in der Praxis regelmäßig entfernt werden muss, um Auswaschungen vorzubeugen. Im Vergleich zu Metallen war die Stickstoffreini-gung weniger effektiv. Warme Temperaturen (bis 20°C) führten zu erhöhter Nitrifikation im Filter. Da keine anoxischen Bedingungen vorhanden waren, fand keine Denitrifikation statt und es wurden große Mengen Nitrat/Nitrit ausgespült, was zu einer Stickstoffnettoproduk-tion führte. Demgegenüber wurde eine effektive Phosphorreinigung erzielt, was durch den hohen Anteil partikulären Phosphors begünstigt wurde. Zu Beginn der Bodenfilteroperation wurde ein mit der Zeit abnehmendes Auswaschen von Feinsediment und Phosphor be-obachtet. Dieses unterstreicht die Bedeutung des Filtermaterials für die Phosphorreinigung, da es (bei hohem Feinsediment- und Phosphorgehalt) zu Phosphorauswaschung aus dem Filter selbst kommen kann.
Biofilter haben das Potenzial, Regenwasser effektiv zu reinigen; in den meisten Fällen lagen die Schadstoffkonzentrationen im gereinigten Wasser deutlich unter denen des Regenwassers. Je nach den lokalen Rahmenbedingungen, den zu entfernenden Schadstoffen und der ökolo-gischen Empfindlichkeit des Vorfluters werden Modifikationen des Bodenfilterdesigns empfohlen. Vor allem hinsichtlich der Stickstoffreinigung und der Anpassung an Winterver-hältnisse besteht weiterer Forschungsbedarf.
ABBREVIATIONS
BMP Best Management Practice
Cd Cadmium
C Carbon source (wood chips and pea straw embedded in biofilter media)
Cu Copper
CV Coefficient of Variance
EPA Environmental Protection Agency
FAWB Facility for Advancing Water Biofiltration
LID Low Impact Development
LTU Luleå University of Technology
N Nitrogen
NH4 Ammonium
NOx Nitrate and nitrite
NTNU Norwegian University of Science and Technology
P Phosphorus
PAH Polycyclic aromatic hydrocarbons
Pb Lead PE Polyethylene
PVC Polyvinyl chloride
RSZ Retrofitted SZ
SZ Water saturated zone / submerged zone
TN Total Nitrogen
TP Total Phosphorus
TSS Total suspended solids
WSUD Water Sensitive Urban Design
APPENDED PAPERS
Paper I.Blecken, G.-T., Zinger, Y., Deletic, A., Fletcher, T.D., Viklander, M.
(2009). Impact of a submerged zone and a carbon source on heavy metal removal in stormwater biofilters. Ecol. Eng. 35 (5), 769-778.
Paper II.Blecken, G.-T., Zinger, Y., Deletic, A., Fletcher, T.D., Viklander,
M. (2010). Effect of retrofitting a saturated zone on the performance of biofiltration for heavy metal removal - preliminary results of a laboratory study. 7th International Conference on Sustainable Techniques and Strategies in Urban Water Management NOVATECH 2010, Lyon, France.
Paper III.Blecken, G.-T., Zinger, Y., Deletic, A., Fletcher, T.D., Viklander,
M. (2009). Influence of intermittent wetting and drying conditions on heavy metal removal by stormwater biofilters. Water Res. 43 (18), 4590-4598.
Paper IV.Blecken, G.-T., Muthanna, T.M., Zinger, Y., Deletic, A., Fletcher,
T.D., Viklander, M. (2007). The influence of temperature on nutrient treatment efficiency in stormwater biofilter systems. Water Sci. Technol. 56 (10), 83-91.
Paper V.Blecken, G.-T., Zinger, Y., Deletic, A., Fletcher, T.D., Hedström,
A., Viklander, M. (submitted). Laboratory study on stormwater biofiltration: nutrient and sediment removal in cold temperatures. Re-submitted to J. Hydrol. after moderate revision, August 2010.
Paper VI.Blecken, G.-T., Marsalek, J., Viklander, M. (submitted). Laboratory
study on stormwater biofiltration in cold temperatures: metal removal and fates. Submitted to Water Air Soil Pollut. September 2010
Paper VII. Muthanna, T.M., Viklander, M. Blecken, G.-T., Thorolfsson, S.T.
(2007). Snowmelt pollutant removal in bioretention areas. Water Res. 41 (18), 4061-4072
Papers I, II and III are based on experiments conducted at Monash University in Melbourne, Australia. For those papers I was responsible for the data interpretation and writing. Furthermore, I participated in the experimental work described in paper III during my stay at Monash University. My co-authors commented on the data interpretation and the paper drafts.
Papers IV, V, and VI are based on experiments conducted at Luleå University of Technology. I significantly contributed to the ideas, the experimental design and work, the data collection and interpretation, and writing the papers, partly with supervision and guidance by Maria Viklander, Tim D. Fletcher, Ana Deletic and Yaron Zinger. Drafts of papers IV, V and VI were also commented upon by Tone M. Muthanna, Annelie Hedström and Jiri Marsalek, respectively. The practical laboratory work was supported by Monica Olofsson and Kerstin Nordqvist.
Paper IV presents and discusses preliminary results of the same study considered in Paper V. However, Paper V presents considerably more results, statistical analyses, kinetic parameters and discussion; thus the two papers do not significantly overlap. The experiment reported in Paper VII was conducted in Trondheim, Norway, at the Norwegian University of Science and Technology. I participated in some of in the experimental work and planning and commented on a draft of the paper.
My contribution to the scientific papers was thus as outlined in the table below. Paper Idea Experimental
design Experimental work Data interpretation Writing I, II responsible responsible
III participation responsible responsible IV, V, VI participation participation responsible responsible responsible
VII minor participation Commenting on paper draft Luleå, September 2010 Godecke-Tobias Blecken
TABLE OF CONTENT
INTRODUCTION 1
1.1 Thesis structure 2
1.2 Background 2
1.2.1 Stormwater quantity and quality 2
1.2.2 Stormwater impact on recipient waters 5
1.2.3 Sustainable urban drainage systems 6
OBJECTIVES 9
BIOFILTER FUNCTIONS AND PROCESSES 11
3.1 Function and design 11
3.2 Pollutant treatment and associated processes 13
3.2.1 Metal and Total Suspended Solids removal 13
3.2.2 Nutrient removal 16
MATERIALS AND METHODS 19
4.1 Biofilter column experiments 19
4.1.1 Column design 19
4.1.2 Stormwater application 22
4.1.3 Design variations 23
4.1.4 Sampling 25
4.2 Biofilter box experiments 26
4.2.1 Biofilter box design 26
4.2.2 Snow application 26
4.2.3 Sampling 27
4.3 Analytical methods 27
RESULTS 29
5.1 Effect of inflow on outflow metal concentrations 31
5.2 Effluent concentrations at different temperatures over run-time 32
5.2.1 Total Suspended Solids 32
5.2.2 Metals 32
5.2.3 Nutrients 34
5.3 Snowmelt pollutant removal 37
5.4 Impact of a submerged zone and a carbon source on pollutant
removal 37
DISCUSSION 47
6.1 Biofilter treatment processes 47
6.1.1 TSS treatment 47 6.1.2 Metal treatment 50 6.1.3 Phosphorus treatment 57 6.1.4 Nitrogen treatment 58 6.2 Biofilter design 60 6.3 Biofilter maintenance 61
6.4 Treatment efficiency and target pollutants 62
CONCLUSIONS 65
FURTHER RESEARCH 67
CHAPTER 1
INTRODUCTION
Urbanisation results in increased stormwater volumes and peak flows as well as impaired water quality. Stormwater contamination is a major cause of ecological degradation of receiving water bodies. Consequently, in recent decades there has been increasing concern about pollutants in stormwater discharges into recipient waters (Ellis and Marsalek 1996). Hence, there is a need to develop, and implement, sustainable stormwater treatment technologies that reliably remove pollutants and are adapted to local ambient conditions.
Besides their capacity to attenuate flows and minimise runoff volumes, stormwater biofilter infiltration systems can improve effluent water quality using vegetated filter media (Davis et al. 2001b), thus contributing to a more sustainable urban environment. Furthermore, biofilters can be integrated into the urban environment, forming an important aesthetically pleasing part of the streetscape (Figure 1).
Figure 1: Stormwater biofilter installation as part of the urban streetscape, Sydney, Australia. Photo courtesy of Professor Tony Wong, Monash University, Australia.
At the beginning of this PhD research project (2005), only a few studies about the pollutant removal efficiency of biofilters had been published (Lau et al. 2000; Mothersill et al. 2000; Davis et al. 2001b; Mazer et al. 2001; Davis et al. 2003; Kim et
al. 2003; Scholz 2004; Lloyd et al. 2001). The increasing popularity of this technology is reflected in the growing number of studies published since then globally, among others the papers appended to this thesis. While the first studies mainly investigated the overall pollutant removal performance of biofilters, subsequently the focus was increasingly on more detailed aspects of biofilter treatment, e.g. their efficiency under specific ambient conditions, the underlying processes and means to enhance their performance. This doctoral work should also be seen in this context; the effects of a range of factors possibly influencing removal efficiencies were investigated and the underlying processes discussed.
1.1 Thesis
structure
This compilation thesis considers the research work presented in seven scientific papers, which are referred to as Papers I to VII (listed above). Four of these papers have been published in international, peer-reviewed scientific journals, two have been submitted to such journals and one presented at an international peer-reviewed conference.
The first chapter, the Introduction, briefly outlines the context of the research. It introduces the research field of urban hydrology and sustainable urban drainage, in which biofilters are used, thus outlining the significance of the research. Based on this, the research objectives are developed. The third chapter, Stormwater biofilter processes and
functions, reviews the relevant literature focusing on water quality improvement by
biofilters. The Materials and Methods chapter presents the experimental set-up and operation as well as the laboratory and data analyses. The Results of these studies are described in the fifth chapter. In the following Discussion those results are compared and discussed in relation to the objectives of the thesis and the relevant scientific literature, and practical implications of this research are summarised. Finally, the main
Conclusions are presented. In the last chapter, recommendations for Further Research are
developed.
1.2 Background
1.2.1
Stormwater quantity and quality
Urbanisation causes significant changes in the hydrological cycle compared to natural and rural catchments in terms of both hydrology and water chemistry and, thus, stormwater runoff has become a significant problem worldwide (Chocat et al. 2001).
Increased catchment imperviousness leads to reduced infiltration and evapo-transpiration of precipitation water, and thus to increased surface runoff (Walsh et al. 2005). Compared to surface runoff hydrographs for natural or rural areas, urban runoff hydrographs (Figure 2) are characterised by a higher total runoff volume, a higher peak flows and a shorter time of concentration, hence they have steeper limbs (Rose and Peters 2001; Roy et al. 2009).
Figure 2: Effect of urbanisation on runoff hydrographs (Butler and Davies 2004).
Depending on the catchment characteristics, urban stormwater might be polluted with a wide range of substances (Ellis and Marsalek 1996). From about 1970 onwards, stormwater pollution has been identified as a significant environmental problem (Malmqvist 1983; Heaney and Huber 1984; Marsalek et al. 1999). Inter alia, significant loads of sediments, heavy metals, nutrients, oils, grease, bacteria and salt pollutants have been found, globally, in urban stormwater and snow (Söderlund and Lehtinen 1971; Browman et al. 1979; Lygren et al. 1984; Hewitt and Rashed 1990; Barret et al. 1998; Viklander 1999; Marsalek 2003; Lee and Bang 2000; Taylor et al. 2005). Stormwater contamination is heavily affected by the density, type and extent of urbanisation as well as the number and pattern of potential contamination sources (Carle et al. 2005).
Suspended solid sources include (inter alia) wet and dry atmospheric deposition, road and tyre wear, construction sites and erosion, and generally dense urban catchments generate higher TSS concentrations than more lightly urbanised areas (Duncan 1999). TSS are of varying sizes (Piro et al. 2010) and especially the fine fractions (<250 m) of suspended solids are carriers of many other (particle-bound) pollutants such as metals, polycyclic aromatic hydrocarbons (PAHs) and phosphorus and are thus frequently used as an easily quantifiable indicator of stormwater pollution (Sansalone and Buchberger 1997a; Lau and Stenstrom 2005).
Phosphorus in stormwater can originate from sources such as construction sites,
atmospheric deposition, fertilisers and plant debris (Carpenter et al. 1998; Duncan 1999) and is often bound to particles (Correll 1999) since it often exists as phosphate ions, which have strong affinity for sediment particles (Berge et al. 1997). Nitrogen is released from fertilisers, animal droppings, plant debris and other organic material, fossil fuel combustion, and wet deposition (Carpenter et al. 1998; Duncan 1999). In urban areas, residential or park catchments produce stormwater with especially high nutrient loads (Clark et al. 2007).
The most commonly reported metals in stormwater are Cd, Cu, Pb and Zn (Duncan 1999); however other metals have also been found (Makepeace 1995). Cd sources include combustion, tyre and brake wear, corrosion of galvanised metals, dumped batteries, fertilisers and pesticides (Makepeace 1995). Cu is released from tyres, engine and break wear, industrial emissions, fungicides and pesticides, leading to especially high concentrations in road runoff and runoff from non-residential urban catchments (Makepeace 1995). Nowadays, common Pb sources include tyre wear, deposition from industrial sources, paints and roof construction materials (Good 1993; Duncan 1999; Makepeace 1995). Before their use was banned, the most important source of Pb were petrol additives. Thus, using data presented in studies published from the 1970s to the early 1990s, Duncan (1999) found that Pb concentrations were highest in road runoff. However, Pb concentrations in road runoff have declined since source control measures were introduced, providing an example of the potential effects of such measures (Fuchs et al. 2002). The main Zn sources are tyre and brake wear, corrosion of galvanised roofs and other building materials (Duncan 1999; Makepeace 1995). Consequently, the highest Zn concentrations can be found in roof and road runoff (Duncan 1999).
The pH of stormwater ranges from 4.1 to 8.3, with a mean of ca. 7 according to Duncan (1999), and thus it is higher than rainfall pH. The cited author found no significant differences reported in stormwater pH among urban catchment types, except that direct roof runoff had a significantly lower pH than other types due to its generally short time of concentration. The pH of stormwater is particularly important for the solubility of a wide range of metals in it, hence the acute toxicity of a range of metals is influenced by pH (Makepeace 1995).
High amounts of salt are applied inter alia as a de-icing agent, especially during winter in areas with cold or temperate climates (Marsalek 2003). Besides its detrimental environmental effect (esp. plant damages; Caraco and Claytor 1997), salt affects the partitioning of metals (especially Cd and Zn) increasing the proportions present in the more environmentally hazardous dissolved phase (Marsalek 2003; Bäckström et al. 2004; Warren and Zimmermann 1994).
Stormwater contamination can vary extremely, both between different storm events and during single events, due to variations in a range of factors, e.g. catchment characteristics, the amount of precipitation, the duration of the preceding dry period,
stormwater flow and season (Duncan 1999; Westerlund et al. 2003; Browman et al. 1979). For instance, Duncan (1999) found total P, Cu and Zn concentrations in urban runoff ranging between 0.01-4.7, 0.005-7.0 and 0.01-43.7 mg L-1 in a review of 306, 192 and 235 records, respectively, and similar variations for other contaminants. Due to factors such as combustion processes being less efficient at cold temperatures, increased heating and accumulations of contaminants in snow packs, pollutant concentrations have been shown to be higher during winter in areas with cold and temperate climates (Engelhard et al. 2007; Westerlund et al. 2003; Hallberg et al. 2007). Within a single runoff event, first flush phenomena may occur (i.e. pollutant concentrations may be highest in initial phases of the storm events; Bertrand-Krajewski et al. 1998).
1.2.2
Stormwater impact on recipient waters
The stormwater drainage system (e.g. proportion of directly connected impervious surfaces, combined or separated sewer system and stormwater treatment facilities) is a major determinant of the impact of stormwater contamination on recipient waters (Taylor et al. 2004; Lee and Bang 2000). Traditionally, stormwater has been managed for flood control, and direct connections between impervious surfaces and recipient waters via separate or combined sewers have been used to facilitate quick and efficient collection, conveyance and discharge of surface runoff water (Figure 3). However, these systems were usually installed without considering water quality aspects (Bedan and Clausen 2009; Walsh et al. 2005). Thus, while there has been substantial success in reducing point source discharges of pollutants (due to increasing implementation of effective waste water treatment) (Swedish EPA 2009), non-point source discharge remains a significant problem (U.S. EPA 2005).
Urban stormwater is a major cause of ecological degradation of urban waterbodies, resulting in impairments of water quality, hydrology and habitats, often collectively referred to as ‘urban stream syndrome’ (Walsh et al. 2005). Common symptoms include increased hydrologic flashiness, higher stream velocities, enlarged stream profiles with modified morphology and eroded riverbanks, reduced base flows, water contamination, reduced biodiversity and higher proportions of tolerant species (U.S. EPA 2002; Walsh et al. 2005; Swedish EPA 2006; Roy et al. 2009; Taylor et al. 2004; Masterson and Bannerman 1994). Urban streams exhibit significantly elevated levels of TSS, heavy metals, nutrients, oil and grease, faecal coliform bacteria and biochemical oxygen demand compared to unimpaired reference streams (Carle et al. 2005; Masterson and Bannerman 1994).
Sediments might accumulate on the bottom of recipient waters, where they may change the aquatic habitat (Masterson and Bannerman 1994) and may be re-suspended in the future. Long-term accumulation might lead to markedly elevated metal concentrations in bottom sediments in urban recipients (Rentz 2008). Cu, Cd and Pb have been shown to bioaccumulate in urban water organisms (Masterson and Bannerman 1994), and stormwater can exhibit acute toxicity (Marsalek et al. 1999). Generally, dissolved metals are more bioavailable than particle-bound metals. Urban non-point sources contribute significantly to nitrogen and phosphorus contamination of surface waters (Carpenter et al. 1998). Even if P is mainly particle-bound it can be hazardous for the environment since it may be released from accumulated (stormwater) sediment (Correll 1999). This nutrient contamination might result in increased algal growth (possibly eutrophication) and oxygen deficiency (Taylor et al. 2004).
1.2.3
Sustainable urban drainage systems
Since urbanisation is increasing rapidly there is an urgent need to develop new, more sustainable solutions to mitigate the negative impacts of traditional urban drainage systems on the natural and urban environment (Chocat et al. 2007; Roy et al. 2009). Therefore, concepts like Water Sensitive Urban Design (WSUD) and Low Impact Development (LID) have been developed in efforts to re-establish near natural water cycles and restore the ecological condition of urban streams (Dietz 2007; Melbourne Water 2005; Walsh et al. 2005; Bedan and Clausen 2009; U.S. EPA 2000).
Figure 4: Urban streetscape before and after retrofitting a sustainable urban drainage systems. Photo courtesy of Kevin R. Perry, Nevue Ngan Assoc., Portland, OR, USA.
While public safety (flood prevention) remains a priority target in these concepts, water quality issues are also taken into account (water quality treatment and source control to improve ecological conditions of urban water bodies). The aim of LID or WSUD is to bring the urban hydrology more in line with pre-development conditions (Bedan and Clausen 2009; Dietz 2007). Furthermore, it is intended to enhance public amenities by integrating stormwater management technologies into the urban streetscape, (Figure 4; Lloyd et al. 2002). Thus, sustainable concepts for new installations, or improvements of existing urban drainage systems focus on: à Stormwater quantity: increased runoff volumes and peak flows in urban areas have
to be reduced.
à Stormwater quality: discharge of stormwater contaminants to recipient water bodies has to be reduced.
à Amenity: stormwater should be made visible to citizens and should be part of the urban streetscape.
In addition to non-structural options (e.g. source control, street sweeping, education), WSUD and LID incorporate a range of often combined retention and treatment technologies, commonly termed structural Best Management Practices (BMP). Examples of commonly implemented BMPs are listed below (based on Melbourne Water 2005; Barret 2005; Deletic and Fletcher 2006; Dietz 2007; Fletcher et al. 2003; Mikkelsen et al. 2001; Pitt et al. 1999; Stahre 2008; Villarreal et al. 2004):
à Buildings can be covered with a vegetated growing media over a waterproof layer to reduce stormwater runoff by disconnecting impervious surfaces from the stormwater pipe system. These green roofs can also provide a source control measure, if they replace metal roofing for instance. In addition, the ‘heat island effect’ is minimised when green roofs are installed.
à Swales convey stormwater relatively slowly (instead of or in combination with pipe systems) using overland flow and gentle slopes. Some sediment removal takes place and thus swales can provide pre-treatment in a stormwater treatment train. à Buffer strips provide sediment removal by routing stormwater as a distributed sheet
flow over a vegetated area.
à In addition, stormwater retention/sediment basins remove TSS by settling sediments. Design criteria are thus both stormwater volume and quality.
à Vertical flow constructed wetlands are artificial vegetated swamplands that remove pollutants by sedimentation, fine filtration and plant uptake.
à Permeable pavements reduce the proportion of impervious surfaces in urban areas, allowing stormwater to infiltrate the in-situ soil and thus recharge groundwater. Pollutants may be trapped or transported to the groundwater. Commonly used materials are concrete blocks or grids, or permeable pavements. Clogging of surfaces is a common problem and pollutant export to the groundwater may be of concern.
à Similarly, infiltration measures (e.g. soakaways, infiltration trenches or basins) capture stormwater and facilitate infiltration and groundwater recharge, thus reducing downstream peak flows and volumes. Since infiltration measures do not treat the water, they may have to be combined with pre-treatment systems.
à The aim of stormwater biofilters (Figure 6; also known as rain gardens or bioretention systems) is to retain stormwater flows and similarly improve water quality using filtration by soil media and vegetation incorporating a range of physical and biochemical processes, as described below.
Bedan et al. (2009) have compared two subdivisions of the same catchment in Connecticut, USA. One subdivision was a traditional and the other a low impact development (the latter incorporating swales, biofilters and permeable pavements). While storm flows and peak discharges from the traditional catchment increased dramatically after development, they were decreased from the LID development. Metal mass exports from the LID catchment were significantly lower than those from the traditional catchment. However, nitrogen and phosphorus concentrations were lower in the runoff from the traditional catchment, possibly due to homeowners’ lawn care practices, fertiliser use on swales and leaching from organic litter. In addition, Rushton (2001) has shown that low impact design for a parking lot led to reductions in runoff volumes, peak flows and pollutant loads.
CHAPTER 2
OBJECTIVES
As with many other WSUD technologies (Marsalek et al. 2003), biofilters have been mostly developed without specific consideration for their operation under varying climate conditions. The biofilter treatment efficiency depends on a range of physical, biological and chemical processes, which to unknown extend might be influenced by variations of the ambient environmental conditions.The overall aim of this thesis has been to provide a scientific basis for achieving reliable target pollutant removal from stormwater by biofilters under given varying conditions (Figure 5).
The effects of various ambient conditions, stormwater characteristics and modifications of filter design on the water quality treatment performance of biofilters has been investigated to: clarify the roles of features of biofilters, elucidate pollutant removal processes inside them, and facilitate the design of effective biofilters.
Adaptation of filter design:
Submerged zone and depth Carbon source Choice of filter material
Ambient conditions:
Temperature variation Length of antecedent drying
Receiving water
Stormwater characteristics:
Target pollutants Partition total - dissolved
Concentration levels
Biofilter processes
Treatment performance Aim: reliable removal of target pollutants
under given ambient conditions due to adapted filter design.
Figure 5: Examples of factors and their interactions (grey arrows) influencing biofilter
The ambient factors investigated were: à temperature,
à length of time-interval between the storm events, The stormwater characteristics investigated were:
à stormwater or snowmelt inflow. à the pollutant concentration.
The filter design modifications investigated were: à coarse filter material,
à a submerged zone, introduced to facilitate de-nitrification,
à an embedded carbon source, introduced to facilitate de-nitrification. The pollutant pathways through biofilters were also investigated.
The main focus has been on the removal of metals and TSS. However, nutrient removal under varying temperatures has also been investigated.
CHAPTER 3
BIOFILTER FUNCTIONS
AND
PROCESSES
Vegetated vertical flow stormwater biofilters (also known as rain gardens or bioretention systems) are one promising system used in WSUD and LID for in situ water quality improvement and flow retention (Davis et al. 2009; Melbourne Water 2005). Biofilters were developed in the USA in the early 1990s (Prince George's County 1993). They are aesthetically pleasing, close-to-nature systems that can be widely applied (or retrofitted) even in dense urban areas (Figure 1 and Figure 6).
Figure 6: A stormwater biofilter as an example of a sustainable urban drainage technology. Photo courtesy of Kevin R. Perry, Nevue Ngan Assoc., Portland, OR, USA.
3.1
Function and design
There are several published guidelines describing the general function and design of stormwater biofilters (e.g. Prince George's County 2007; Melbourne Water 2005; U.S. EPA 2004), according to which they typically consist of a vegetated swale or basin, underlain by a filter medium (Figure 7). A detention storage on top allows temporary ponding of water if the inflow exceeds the infiltration flow (Figure 8). Stormwater is conveyed to the biofilter through an inlet with erosion protection (Figure 6). The area of a biofilter is commonly between 2 and 5% of the impervious catchment area.
The filter itself, which is usually ca. 700-900 mm thick, consists of either natural soil or engineered media (sandy loams commonly are used), in some cases topped by a 5-10 cm thick mulch layer (Melbourne Water 2005; Prince George's County 2007). Stormwater infiltrates and percolates through the filter and during its passage it is filtered by the filter media, plants and microbes via a combination of mechanical and biochemical processes (Davis et al. 2001b). Vegetation was furthermore shown to be critical in maintaining the infiltration capacity of biofiltration systems (Lewis et al. 2008). The treated water is either infiltrated into the surrounding soil or collected in a drainage pipe at the bottom of the filter and then discharged to a recipient or the existing sewer system. Biofilters can be conveniently retrofitted even in dense urban developments (Smith et al. 2007).
Figure 7: Cross-sectional drawing of a stormwater biofilter equipped with an underdrain
Figure 8: Biofilter during a storm event with stormwater ponding on top of the filter. Photo courtesy of Kevin R. Perry, Nevue Ngan Assoc., Portland, OR, USA.
3.2
Pollutant treatment and associated processes
From about 2000 onwards, a number of studies have been published on the pollutant removal performance of biofilters; a summary of metals and nutrient inflow and outflow concentrations observed in selected studies is given in Table 1 and Table 2. Removal of other pollutants such as oil/grease (Hong et al. 2006), PAHs (Diblasi et al. 2009) and bacteria (Garbrecht et al. 2009; Hathaway et al. 2009; Rusciano and Obropta 2007) by biofilters have been investigated in a few studies, but are not discussed in detail here since they were not pollutants of concern in this thesis.
3.2.1
Metal and Total Suspended Solids removal
Total metal and TSS removal by stormwater biofilters often exceeds 80-90% (Bratieres et al. 2008; Davis et al. 2001b; Davis et al. 2003; Hatt et al. 2009; Hsieh and Davis 2005b; Lau et al. 2000; Muthanna et al. 2007; Read et al. 2008; Sun and Davis 2007). Since most metals entering biofilters are particle-bound, mechanical filtration of the incoming stormwater sediment also removes substantial loads of metals (and other particle-bound pollutants), thus the efficiency of TSS and particle-bound metal removal is correlated (Hatt et al. 2008).
In most biofilter studies only the total metal removal has been investigated (see Table 1), dissolved metal removal has been considered in fewer of the investigations (Chapman and Horner 2010; Lau et al. 2000; Muthanna et al. 2007; Read et al. 2008; Hatt et al. 2007; Sun and Davis 2007). Some findings indicate that dissolved metal removal is significantly lower than total metal removal, in particular Cu leaching being observed (Hatt et al. 2007; Chapman and Horner 2010; Muthanna et al. 2007). However, biofilters seem to have clear potential to provide adequate dissolved metal treatment (Sun and Davis 2007; Hsieh and Davis 2005a), while dissolved metals are not targeted at all e.g. in sedimentation ponds, although dissolved metal removal is of special concern since dissolved metals are far more bioavailable than particle-bound metals (Morrison 1989).
Dissolved metal removal is affected by diverse factors that influence soil and heavy metal interactions. Numerous studies in the field of soil sciences have investigated these interactions in soils. Dissolved metals may be removed by biofilters via sorption processes in the filter (cation exchange, specific adsorption, precipitation and organic complexation; Rieuwerts et al. 1998) and/or plant uptake (Fritioff and Greger 2003).
14 : B io fil te r infl ow a nd out flo w c onc en tr at ions of TS S ( m g L -1 ) a nd meta ls ( g L -1 ) fro m selected s tu dies . Data
partly retrieved from Davis
et al. ( 2009). ., lab orato ry study; d iss., diss ol ved con cen trations; n on-ve g., n on-vegetated; In. , inflow concentration s; Out.: ou tfl ow co nc en trat ions) TS S Cd C u Pb Zn Filter type Desi gn att ibutes To tal / d iss. In Out In Out In Out In Out In Out Reference La b. 150 (1 07 )-<6 H sieh et a l. (2005a) La b. N o n-veg. 42 21 22 7 41 3 285 122 B irch et al. ( 2005) La b. N o n-veg. 20 6 6 23 7 6 14 6 <1 1 8 00 10 R e ad et al. (20 08 ) Veg etat e d 20 6 5 23 7 5 14 6 <1 1 8 00 20 P ilo t 14 0 3.4 61 <2 60 0 <2 5 D a v is et al. (20 01b ) 64 4. 9 <2 <2 59 0 <2 5 Pilo t 38-160 2.3-5.2 32-104 <2 260 -1 2 90 < 2 5 Dav is et al. ( 2 00 3) P ilo t Su mmer Total 12 6 41.7 21.1 2.5 58 4 49 Summer Diss. <10 44 1.4 1.1 Wi nt e r Total 26 15.2 4.4 0.8 41 2 22 Muthan na et al. (200 7) Wi nt e r Diss. <1 0 10 0.05 0.02 6 Fi eld 5. 6 1. 9 56. 8 1. 9 41. 4 10 .2 98. 3 20. 6 G lass et al. (20 05 ) Fi eld 34 13-1 8 10 7 44-4 8 D a v is (2 00 7 ) Fi eld 50 20 72 17 H u n t et al. (20 08 ) Fi eld 39 3 to 5 10 4 to 6 6 2-3 10 0 13-3 0 H a tt et al. (2 0 0 9 ) 12 8 14 5 <1 60 5 11 0 7 33 0 13 Fi eld 66 6 19 16 6 3 71 12 Li et a l. (20 09 ) 14 4 13 9 <2 <2 15 3 Fi eld Total 12 0 30 16 6.3 17 4.5 12 0 47 Chapman et al . (201 0 ) Diss. 3. 6 2. 9 <1 <1 49 26
15 : Biofilter i nflo w a nd o utflo w con cen tr at ions of nutrients ( m g L -1) fro m selected s tu dies . Data partly retrieve d from Davis et al. (2009). P N Nitrit e/n it rat e -N Am m o nia -N Desi gn att ibutes To tal / d iss. In Out In Out In Out In Out Reference .
Partly with satur
a ted zon e 2. 1 0. 1-3 K im et a l. (2 00 3) 0.25 0.12 2. 14 1.81 0.62 1.34 B irch et al. ( 2 0 0 5 ) 3 0.46-2.9 H sieh et a l. (2005 a) e getat e d 0.48 0 .03-0.0 7 5. 44 1.23-2.0 4 0.69 0.05-0. 2 4 0.48 0. 02-0.0 3 H e nderson et al. (20 07 ) Non-veg. 0. 48 0 .05-0. 3 3 5. 44 4. 06-6. 0 9 0. 69 1. 75-2. 6 9 To tal 0.26 0 0.10 6 1.02 2.18 0.39 3 0.3 7 0 0.1 1 3 0.47 4 R e ad et al. (20 08 ) Non-veg. Diss. 0. 03 3 0. 08 0. 65 1. 86 V e getat e d To tal 0. 26 0 0. 08 2 1. 02 1. 69 0. 39 3 0. 0 8 3 0. 11 3 0. 45 6 V e getat e d Diss 0. 03 3 0. 02 7 0. 65 1. 40 lot 0.44 0.13 1. 2 0.48 D a v is et al. (20 01b ) 0. 52 0. 1 0. 34 0. 26 2. 4 0. 5 lot 0.28-0.8 8 0 .06-0.1 5 1.6-6 1. 1-2.8 D a v is et al. (20 06 ) d 0. 01 2-0. 01 9 0. 05 8-0. 06 Di etz et al. (20 06 ) d 0. 11 0. 56 1. 35 4. 38 H u n t et al. (20 06 ) d 0.61 0 .15-0.1 7 D a v is (2 00 7 ) d 0. 19 0. 13 1. 68 1. 14 H u n t et al. (20 08 ) d 0. 07 0 .16-0. 2 2 1. 1 1. 1-1. 3 0. 4 0. 14 -0. 3 0. 04 0. 02-0. 0 3 H a tt el at. ( 2 0 0 9 ) 0. 4 0. 07 2. 7 2 .2 1 1. 6 0. 5 0. 02 d 0. 1 0. 35 1. 6 2 .5 0. 36 1. 0 Li et a l. (20 09 ) <0. 1 <0. 1 1 0 .6 0. 34 0. 05 d 0.21 0.13 1.15 0.81 Chapman et a l. (201 0 )
Of particular concern for the treatment efficiency of dissolved metals are the filter media and soil solution pH (since pH is the primary determinant of metal solubility and thus metal sorption processes, as is the stormwater pH for the distribution between particle-bound and dissolved metals in stormwater; Sansalone and Buchberger 1997b; Rieuwerts et al. 1998; Bradl 2004). However, in addition to pH, a range of other physical and (bio)chemical factors may influence metal solubility and soil surface chemistry (Bradl 2004; Warren and Haack 2001). Notably, the redox status also affects the partitioning of elements between solution and solid phases (Rieuwerts et al. 1998; Warren and Haack 2001).
As in stormwater, increasing salinity may raise the solubility of metals and thus reduce rates of metal sorption in the filter (Förstner et al. 1998; Warren and Zimmermann 1994). Temperature changes may influence biological and biochemical processes (e.g. nutrient treatment and plant metal uptake, see below), but not necessarily influence metal sorption by soils (Lucan Bouche et al. 1997; Rieuwerts et al. 1998).
Plants, especially their root activites, also strongly influence metal solubility in soils, and thus in filter media, via their effects on dissolved organic matter contents and/or pH (Zhao et al. 2007). Furthermore, plants have been shown to take up dissolved metals (Fritioff and Greger 2003), thereby accounting for up to ca. 10% of the total metal removal by biofilters (Davis et al. 2001b; Sun and Davis 2007). However, even non-vegetated soil filters remove metals quite effectively (Hatt et al. 2008; Read et al. 2008). Furthermore, dissolved metals may also be removed by sorption to bacteria and biofilms (Warren and Haack 2001).
Typically, metals do not ingress far into the filter, but are trapped on or near the top of the filter due to both mechanical removal and sorption processes (e.g. Davis et al. 2001b). However, the accumulation of fine stormwater sediment on top of the filter material and in the upper layers reduces the hydraulic conductivity relatively quickly, possibly leading to clogging (Le Coustumer et al. 2007; Li and Davis 2008). On the other hand, the high metal removal in the upper layer facilitates filter maintenance since merely scraping off the top layer may remove a high proportion of accumulated metals from the system, and thus postpone the need to replace the whole filter media (Hatt et al. 2008).
3.2.2 Nutrient
removal
Nutrient removal is far more variable than metal and TSS removal in biofilters, thus nutrient treatment in biofilters might be problematic (Dietz 2007). Notably, both very efficient removal of N and P, and excessive leaching from them, have been observed. For example, Davis et al. (2006) reportedly observed 70-80 % TP removal in biofilter box experiments, but Li and Davis (2009) observed strong leaching (0.1 and 0.35 mg L-1 TP in influent and effluent, respectively). Similarly, the efficacy of total nitrogen treatment is very variable, ranging from effective removal to significant leaching (Kim
et al. 2003). However, although the efficiency of biofilter treatment for both nutrients is similarly variable, the underlying mechanisms differ.
The removal of the mainly particle-bound incoming P is effective due to the mechanical filtration of particulate P (and hence correlated to TSS removal) and sorption of dissolved P by the filter media (Henderson et al. 2007; Hsieh and Davis 2005a; Hsieh et al. 2007a). However, in a number of studies, net P leaching from the filter media has been observed due to wash-out of fine materials with associated P (Hatt et al. 2009; Hunt et al. 2006; Li and Davis 2009; Read et al. 2008), especially from newly constructed filters often decreasing with time due to media stabilisation (e.g. repacking, settling) and/or depletion of the reserves (Hsieh et al. 2007a). Thus, to achieve low P concentrations in the effluent, it is essential to select appropriate filter media (Hunt et al. 2006), and filter media with high P concentrations should be avoided (Dietz 2007; FAWB 2008). In addition to stormwater runoff, eroded sediments are important non-point sources of P (Brady and Weil 2002), thus biofilters might indirectly reduce P discharge to recipients since they reduce surface runoff flows by reducing erosion losses in urban catchments.
Figure 9: Sectional drawing of a stormwater biofilter equipped with submerged zone Net N removal depends on the removal balance of the different N species. Effective NH4 and total Kjeldahl N removal has commonly been shown (Bratieres et al. 2008; Davis et al. 2001a; Davis et al. 2006; Hsieh et al. 2007b; Henderson et al. 2007). However, NOx removal is often inadequate and has been identified as the main reason for the N net leaching commonly observed (Birch et al. 2005; Bratieres et al. 2008; Davis et al. 2001b; Davis et al. 2006; Hatt et al. 2009; Hsieh et al. 2007b; Li and Davis 2009; Kim et al. 2003). To enhance total N removal a submerged zone (combined with a carbon source) has recently been introduced in the filter media to enable denitrification due to (partly) anoxic conditions and thus increase overall N treatment (Figure 9; Kim et al. 2003; Dietz and Clausen 2006).
Vegetated biofilters significantly enhance both P and N removal (Lucas and Greenway 2008; Read et al. 2008), but significant variations in nutrient removal between filters
CHAPTER 4
MATERIALS AND METHODS
Papers I to VI present results of laboratory trials using biofilter columns (mesocosms) with essentially the same design. The studies described in Papers I, II, and III were conducted in the FAWB (Facility for Advancing Water Biofiltration) laboratories at Monash University, Clayton Campus, Melbourne, Australia and those described in Papers IV, V, and VI at Luleå University of Technology (LTU) in Luleå, Sweden. The design of the biofilters used in these studies was based on guidelines published by Melbourne Water (2005). To investigate the aspects considered in the papers, several design modifications of the columns were made, as described below.
The study presented in Paper VII was conducted at the Norwegian University of Science and Technology (NTNU), Trondheim. In this study a different biofilter design was implemented, a so-called biofilter box.
In the following sections the experimental set ups used in each of the studies are summarised. Full details can be found in the papers.
4.1 Biofilter
column
experiments
4.1.1 Column
design
Biofilter columns (Figure 10 and Figure 13), 900 mm tall with an inner diameter of 377 mm (LTU) or 375 mm (FAWB) and an approximate cross-sectional area of 0.11 m2 were made of PVC pipes. The inner walls were sandblasted to increase surface roughness in order to reduce the risk of preferential flow along the columns’ walls. The columns had a 400 mm high transparent pipe on top, which allowed stormwater ponding without shading the plants and thus affecting their growth.
Figure 10: Biofilter column: (a) configuration with riser pipe, (b) general filter media set-up
and vegetation. Figure source: Paper I.
Figure 11: Biofilter columns at LTU under construction: (a) drainage pipe at the filter bottom,
(b) bottom sand layer, (c) top layer with admixed top soil and plants.
The filter itself was 800 mm deep, divided into two 400 mm layers. The specifications of the filter media in those layers is summarised in Table 3, cf. Figure 11 and Figure 12. Since a coarse filter media is recommended for cold climate biofilters (Caraco and Claytor 1997) in order to facilitate sufficient infiltration even into frozen soils (Kane 1980), the filter media used in the experiments at LTU was coarser compared to the media used at FAWB. The filter itself was underlain by a 100 mm deep drainage layer consisting of a transition layer (30 mm) and the drain itself (70 mm). A slotted horizontal drainage pipe (50 mm diameter at FAWB, 58 mm at LTU, Figure 11a) was placed on the bottom of the filter to convey the water to a sampling port. The columns were planted with native Carex species: C. apressa R.Br. (tall sedge) at FAWB (Figure 13a and b) and C. rostrata Stokes (bottle sedge) at LTU (Figure 11c).
Figure 12: Filter media used in the FAWB columns: (a) fine sand, 0.25-0.50 mm, in the
bottom filter layer; (b) sandy loam in the top filter layer; (c) fine gravel, 2-4mm, in the underdrain; (d) wood chips and (e) pea straw as carbon sources.
Table 3: Summary of experimental design specifications of the biofilter columns used in the
studies at Luleå University of Technology (LTU) and Monash University (Facility for Advancing Water Biofiltration - FAWB) presented in Papers I to VI.
LTU columns FAWB columns Column specifications Material PVC PVC Height 900 mm + 400 mm transparent top 900 mm + 400 mm transparent top Riser pipe to create SZ no yes
Sampling pipes in
filter media no
yes; enabling SZ heights of 150, 450, 600 mm
Vegetation
Plant species C. rostrata Stokes
C. appressa R.Br.
(Paper II also Dianella revoluta, and Microleana stipoides) Filter media
Top filter layer (400 mm)
sand with 5% silt and 14% fine gravel,
d50 = 620 μm UC = 2.4
mixed with top soil in the upper 100 mm
sandy loam d50 = 330 μm UC = 6.4
Bottom filter layer (400 mm) fine sand
fine sand *;
partly with embedded carbon source: pea straw/wood chips Underdrain (30 + 70 mm)
coarse sand
fine gravel with embedded drainage pipe
coarse sand
gravel with embedded drainage pipe
Stormwater application
Stormwater volume 15 L twice weekly 25 L twice weekly with design variations Main design variations
Temperature :
5 replicates each in climate rooms at 2, 8 and 20°C
Impact of a (retrofitted) SZ, Impact of C,
Drying and wetting
* The biofilters with the retrofitted SZ described in Paper II had only one filter layer of sandy loam
After constructing, filling and planting the biofilter columns, the biofilters were irrigated with tap water to allow plant establishment and then flushed with stormwater pond water (FAWB) or natural water from an enclosed urban bay in Luleå (LTU) three times during one day each to promote natural biofilm development.
Figure 13: Biofilter columns in a greenhouse, FAWB (a and b) and a climate room, LTU (c).
The biofilter columns at FAWB were placed in a greenhouse (Figure 13a and b) with open mesh on its sides to ensure that local climatic conditions were maintained. At LTU, the columns were placed indoors in climate rooms (Figure 13c). To enable plant growth, they were illuminated with greenhouse lamps 12 h daily. Both arrangements ensured that the columns received inflow water solely by controlled dosing (i.e. non from precipitation).
4.1.2 Stormwater
application
For stormwater dosing, artificial stormwater was utilised: natural stormwater sediment (from a stormwater gully pot for the experiments at LTU and from a stormwater pond for the experiments at FAWB) and laboratory-grade chemicals were added to tap water in amounts required to achieve target stormwater pollutant concentrations (Table 4).
In general, the biofilter columns were dosed twice weekly with 15 L (136 mm) and 25 L (227 mm) stormwater at LTU and FAWB, respectively. The calculation of the stormwater volume was based on typical percentage biofilter area of an impervious catchment with typical rainfall patterns and depths, as described in detail in the Papers I and IV. In between the storm the biofilter was exposed to gravity drainage.
Over the experimental run times, composite samples were taken to enable monitoring of the stormwater inflow quality.
Table 4: Stormwater inflow characteristics for each of the studies (Papers I to VII). Metal
concentrations in g L-1, TSS, chloride and nutrient concentrations in mg L-1.
Pollutant I II III IV V, VI VII
TSS 150 179 143 131 2900 -- 9600 Total Cd 5.8 6.7 4.2 0.30 0.44 1.52 Dissolved Cd 3.4 0.02 0.02 0.01 Total Cu 54.7 67.5 95.1 153.0 123.4 200.4 543.3 Dissolved Cu 21.9 36.4 19.0 13.5 Total Pb 142.9 154.5 181.5 41.2 24.6 34.8 93.7 Dissolved Pb 0.3 0.3 0.18 0.19 Total Zn 576.9 450.3 587.3 267.0 386.4 537.6 1465.3 Dissolved Zn 211.0 7.2 6.1 2.6 Chloride 49.0 215.8 20.9 Total N 2.24 2.84 1.38 1.80 Dissolved N 1.16 1.68 Dissolved NOx-N 0.24 0.4 Dissolved NH4-N 0.32 0.22 Total P 0.47 0.44 0.29 0.31 Dissolved P 0.03 0.03 pH 7.1 6.9 6.9 6.7 – 7.0
4.1.3 Design
variations
The following text describes the design variations of the columns made to enable evaluation of the effects of the factors described in the Papers I to VI. For each design variation at least three (FAWB) or five (LTU) replicates were used to enable statistical analyses and validate the results. In all experiments, the columns were randomly allocated to the different design variation groups.
Percolating water sampling pipes (I). In the filter media, gravel filled PVC sampling
pipes with 21 mm diameter were installed at different depth (Figure 10a). These pipes traversed the entire column diameter, they had open tops to enable sampling of the seepage and were surrounded by sand to prevent washout of the overlying soil. Unfortunately, some clogged, so samples were not available from all depths.
Submerged zone (I, II, III) and Carbon source (I, III). To test the effect of submerged
zones with different heights, the sampling port was elevated using an external riser pipe with six outlet taps installed at different heights (Figure 10 and Figure 13). By opening one of these taps, the water level could be kept constant at the bottom of the column, thus creating submerged zones of different depths (according to the height of the open tap). For the studies presented in Papers I and III, the SZ was fitted to the biofilters from the beginning of the experimental run time, while for the columns investigated in described in Paper II the SZ was retrofitted after one year of biofilter operation as standard filters.
To promote denitrification a cellulose-based carbon source was embedded in the bottom filter layer. For the filters used in the studies presented in Papers I and III, a carbon source was embedded in the deeper filter layer (i.e. in the submerged zone) from the beginning. No additional carbon source was implemented if the submerged zone was retrofitted into standard biofilters (cf. Paper II). The carbon source consisted of 1/3 pea straw and 2/3 wood chips. Thus, in the studies presented in Papers I to III, different combinations of SZ and C were tested as outlined in Table 5.
Table 5: SZ and C combinations in the studies presented in Papers I to III. RSZ: Retrofitted SZ after ca. one year of operation. SZ height in mm,cf. Figure 10.
Paper I Paper II Paper III
SZ C SZ C SZ C 0 non 0 non 0 non
450 non RSZ 450 non 450 with
0 with
150 with
450 with
600 with
Extended drying periods (III). To investigate the effects of extended drying on the
filter performance (Paper III), the columns were divided into three groups of six columns each: one of biofilters equipped with a 450 mm SZ combined with C and one of standard biofilter columns. The three groups were dosed with stormwater according to three different drying and wetting schemes (Table 6).
Table 6: Drying and wetting scheme, cf. Paper III.
Group Wet Dry Wet Dry Wet Dry Wet
A 1 week 7 weeks 4 weeks
B 1 week 1 week 1 week 2 weeks 2 weeks 3 weeks 2 weeks C 1 week 4 weeks 2 weeks 3 weeks 2 weeks
During wet periods, the columns were dosed with 25 L stormwater twice weekly while they received no inflow at all during dry periods. As for the whole biofilter system, the submerged zone was also subject to drying, and thus allowed to draw down over time. In this manner, the experimental operation realistically reflected the behaviour of a real biofiltration system during long dry periods.
Temperature (IV, V, VI). To investigate the effect of varying temperatures on biofilter
treatment efficiency under controlled laboratory conditions, the columns were tested in three thermostat-controlled climate rooms at nearly constant air temperatures (mean 1.8, 7.3 and 20.3 °C, hereafter referred to as 2, 7 and 20°C(1)). Five columns were placed in every room. The air temperature in the rooms was logged at 15 minute intervals. All columns were illuminated with high pressure sodium greenhouse lamps (400 W, Figure 13c) for 12 hours daily. For these experiments, the stormwater
(1) In Papers IV and VI the intermediate temperature is referred to as 8°C since this was the primary
had to be prepared separately in each room for temperature adjustment, however, no significant differences in stormwater quality between the three temperatures were detected (One-way ANOVA, Į=0.05).
4.1.4 Sampling
Effluent. During the periodic outflow sampling runs composite samples from all
columns were taken. The sampling strategy differed partly between the experiments. In the C/SZ study (Paper I), two samples were taken and analysed in order to evaluate the first (i.e. resident pore water) and later flow (i.e. the newly treated water) from the filter separately. In the studies described in Papers II and III, for each column one 1 L composite inflow sample was collected, consisting of five 200 mL sub-samples. The first sub-sample was collected after 1 L of the treated water had flowed out, followed by four other sub-samples each after a further 5 L outflow of water. In the temperature studies (Papers IV to VI), the outflow water was collected until at least 14 L of treated water had been discharged. This collected water was stirred and then the samples were taken.
Percolating water. In the C/SZ study (Paper I), in addition to the outflow samples
from the underdrain, samples were taken from the lateral outlets (one from each outlet at the start of the sampling event to track the waterfront progress).
Filter media. After the stormwater applications, filter media samples were taken from
the FAWB columns described in Papers I and III and from the columns at LTU. At FAWB, 25 mm thick filter media slices were taken at depths of 0–25, 25–50, 60–80, 200–225, 470–495, and 600–625 mm from each of the columns with and without SZ. The samples were extracted from two vertical cores of 100 mm diameter per column. At LTU, filter media samples were taken at three filter depths: the top (0-50 mm) and at 250 and 600 mm. The top samples were taken manually as grab samples, while the two other samples were horizontal cores (35 mm) spanning the whole column diameter. In addition, for all columns prior to the experiments, a grab sample of the different filter material layers (only the top layer at FAWB) were taken to determine the initial metal concentrations.
Plants. In the experiments at LTU, before and after stormwater application plant root
and shoot samples were collected. Prior to the experiments, a composite sample of randomly selected plants was taken to investigate the initial plant metal concentrations. After finished stormwater application, a composite root and shoot sample was collected separately from each column.
